Boundary-Aware Network for Fast and High-Accuracy Portrait Segmentation
It addresses the need for fast and accurate portrait segmentation, which is incremental as it builds on existing segmentation methods.
The paper tackles portrait segmentation by proposing a lightweight Boundary-Aware Network (BANet) that achieves high-quality segmentation with real-time speed (>25 FPS) and produces finer results than annotations.
Compared with other semantic segmentation tasks, portrait segmentation requires both higher precision and faster inference speed. However, this problem has not been well studied in previous works. In this paper, we propose a lightweight network architecture, called Boundary-Aware Network (BANet) which selectively extracts detail information in boundary area to make high-quality segmentation output with real-time( >25FPS) speed. In addition, we design a new loss function called refine loss which supervises the network with image level gradient information. Our model is able to produce finer segmentation results which has richer details than annotations.